● Software Engineer at amazon working on Trainium AI/ML team.● Trained and fine-tuned large language models and imaging models using Pytorch and AWS Neuron, as well as optimized parameters for Trn1 and Nvidia p4d clusters as a Machine Learning Engineer, contributing to a 50% lowered training cost for equal or better performance to comparable EC2 instances● Developed automated pipelines and scripts for testing and benchmarking models on Trn1, capturing metrics such as compilation/training time, loss curve, and MFU (model flops utilization)● Connected test results to a database to test for regressions and improvements between versions● Produced tutorials on running models for customers on the AWS neuron samples Github(https://github.com/aws-neuron/aws-neuron-samples/tree/master), such as facebook/bart-large for customers such as Databricks, Ricoh, NinjaTech AI, and Arcee AI

Ensure the testability and manufacturability of solid state drives through production ramp.Make significant contributions to design, development and validation of drives.Develops and debugs complex software programs to convert design validation vectors and drive complex test equipment.Creates and tests validation and production test hardware solutions.

Led a 5-person team to build a discord bot from scratch with discord.py.Bot functionalities include conducting office hours, playing a coin flip game, generating Reddit posts, and working heavily with reading and processing user messages.

Worked under chief of ground and programming Robert Winig.Worked with charts.js, java script, and 3D modeling on internal demos.Final project involved creating graphs to show satellite bandwidth using charts.js.